AskScholar

• 

  AskScholar application

Inspiration

I am a student, who is currently trying to finance my studies with the help of scholarships. There are many websites which provide the details of various scholarships. But it is very difficult to look each scholarship in detail and check whether it suits you. I have wondered like if we could get the suitable scholarship, and enquire details about these to someone. This made me to build AskScholar, a question answering chatbot and a friendly scholarship advisor.

What it does

AskScholar is a chatbot application designed to help students discover scholarships and provide answers to queries related to scholarships. The app leverages advanced AI technologies to provide relevant, up-to-date information in a conversational format.

Key Features

• Scholarship Discovery: Easily search for scholarships based on your profile and requirements.
• AI-Powered Query Handling: AskScholar can handle questions related to eligibility, deadlines, application processes, and more.
• Embeddings and Text Generation: Utilizes Gemini for generating embeddings and text responses.
• Retrieval-Augmented Generation (RAG): Powered by TiDB Serverless and TiDB Vector Search for efficient retrieval and generation. ## How we built it
• User Query Input: Users can input their queries or search for scholarships using the chatbot interface.
• Query Processing: The query is processed using LangChain and embeddings are generated using Gemini.
• RAG Implementation: The system uses TiDB Vector Search to retrieve relevant documents, which are combined with text generation to deliver accurate and contextual responses.
• Response Display: The final response is presented through the Streamlit interface. ## Challenges we ran into At first, it was difficult to connect to the cluster. But with the support of the discord community, this challenge was resolved. ## Accomplishments that we're proud of A working chatbot to enquire about scholarships, application process, and other details. ## What we learned
• TiDB Vector Search and TiDB Serverless cluster was a perfect tool to build a RAG for a chatbot.
• Streamlit makes it easy to make AI applications.
• Langchain integrations were very useful. ## What's next for AskScholar
• Add support for more regions and scholarship sources.
• Integrate additional language models for improved accuracy.
• Enhance the UI for better user experience.

Built With

• gemini
• langchain
• python
• streamlit
• tidb